Generate an anova object. The method to calculate the BIC is that from the saemix package. As in other prominent anova methods, models are sorted by number of parameters, and the tests (if requested) are always relative to the model on the previous line.

# S3 method for saem.mmkin
anova(
  object,
  ...,
  method = c("is", "lin", "gq"),
  test = FALSE,
  model.names = NULL
)

Arguments

object

An saem.mmkin object

...

further such objects

method

Method for likelihood calculation: "is" (importance sampling), "lin" (linear approximation), or "gq" (Gaussian quadrature). Passed to saemix::logLik.SaemixObject

test

Should a likelihood ratio test be performed? If TRUE, the alternative models are tested against the first model. Should only be done for nested models.

model.names

Optional character vector of model names

Value

an "anova" data frame; the traditional (S3) result of anova()